Dhakirah Kehinde Salahudin-Mukeem, a rising researcher in Big Data Analytics, has unveiled innovative findings in her Master’s dissertation titled “Hybrid Data Mining Technique for Credit Card Fraud Detection.” Her innovative approach addresses one of the most pervasive challenges in the financial sector: the rising tide of credit card fraud.
[ad]
With over 15 years of professional experience and a strong academic background, Dhakirah combines practical expertise with innovative research. Having recently completed an MSc in Big Data Analytics at Sheffield Hallam University, she specializes in hybrid data mining techniques for credit card fraud detection. Her technical proficiency spans SQL, Python, Power BI, and Tableau, along with expertise in Oracle Database Administration. Dhakirah’s achievements include leading data-driven process optimizations in the aviation sector and developing advanced analytics solutions to enhance operational efficiency and service delivery. This unique blend of skills positions her as a trailblazer in fraud prevention strategies.

With global financial losses due to fraud projected to reach $397 billion over the next decade, Dhakirah’s work at Sheffield Hallam University could not come at a more critical time. Her research explores the limitations of traditional data mining techniques and introduces a hybrid model designed to detect fraud with unprecedented accuracy and adaptability.

Utilizing a simulated credit card transaction dataset of over 1.8 million records, Dhakirah’s hybrid model integrates multiple machine learning algorithms, including Light Gradient Boosting Machine (LGBM), eXtreme Gradient Boosting (XGBoost), and Neural Networks. The results demonstrated that the hybrid model outperformed individual algorithms in key metrics such as accuracy, precision, recall, and Area Under the Receiver Operating Characteristic (AUROC) curve. Notably, combinations involving LGBM consistently achieved the highest predictive performance, setting a new standard for fraud detection.

One of the key findings of her research is the model’s ability to address the challenges of imbalanced datasets, a common issue in fraud detection. Using advanced techniques like SMOTE-ENN (Synthetic Minority Over-Sampling Technique with Edited Nearest Neighbors), Dhakirah effectively balanced the dataset, ensuring that fraudulent and legitimate transactions were equally represented. This resulted in a significant reduction in false positives and negatives, a critical factor in maintaining consumer trust and operational efficiency.

The study also revealed interesting insights into fraud patterns. For example, fraudulent transactions were most prevalent in merchant categories such as online shopping and grocery stores, while younger and middle-aged demographics were more likely to be targeted than traditionally assumed older populations. These findings challenge existing stereotypes and emphasize the importance of data-driven decision-making in designing fraud prevention strategies.

Dhakirah actively prioritized user experience, incorporating feedback from end-users to ensure the system’s practical applicability. Her hybrid model was praised for its ease of use and interpretability, addressing regulatory demands for transparency in artificial intelligence systems. As Dhakirah explains, “Incorporating user-friendly design with cutting-edge technology is key to building trust, both for customers and financial institutions. Fraud detection systems must be as transparent as they are accurate.”

Her dissertation also highlighted potential areas for improvement, including expanding the model to real-world datasets and refining interpretability to meet stringent regulatory requirements. These recommendations offer a roadmap for future researchers and financial institutions seeking to implement hybrid models at scale.

Dhakirah’s work has significant implications beyond academia. Her research provides a pathway for banks, payment processors, and e-commerce platforms to protect financial systems, build customer trust, and minimize operational disruptions from false fraud alerts. She delivers a scalable and adaptable solution that could transforms how financial institutions combat fraud.
[ad]